12913: Apply secondaryFiles to inputs to RunInSingleContainer

[arvados.git] / sdk / cwl / arvados_cwl / runner.py

diff --git a/sdk/cwl/arvados_cwl/runner.py b/sdk/cwl/arvados_cwl/runner.py
index e76d6283f5cf925f965e0b98a33743569685982c..28de7f368a23ccbc52a0cae37fa55d358744d0e1 100644 (file)
--- a/sdk/cwl/arvados_cwl/runner.py
+++ b/sdk/cwl/arvados_cwl/runner.py
@@ -46,11 +46,16 @@ def trim_anonymous_location(obj):
     if obj.get("location", "").startswith("_:"):
         del obj["location"]
 
+def remove_redundant_fields(obj):
+    for field in ("path", "nameext", "nameroot", "dirname"):
+        if field in obj:
+            del obj[field]
+
 def find_defaults(d, op):
     if isinstance(d, list):
         for i in d:
             find_defaults(i, op)
-    if isinstance(d, dict):
+    elif isinstance(d, dict):
         if "default" in d:
             op(d)
         else:
@@ -161,6 +166,8 @@ def upload_docker(arvrunner, tool):
                 raise SourceLine(docker_req, "dockerOutputDirectory", UnsupportedRequirement).makeError(
                     "Option 'dockerOutputDirectory' of DockerRequirement not supported.")
             arv_docker_get_image(arvrunner.api, docker_req, True, arvrunner.project_uuid)
+        else:
+            arv_docker_get_image(arvrunner.api, {"dockerPull": "arvados/jobs"}, True, arvrunner.project_uuid)
     elif isinstance(tool, cwltool.workflow.Workflow):
         for s in tool.steps:
             upload_docker(arvrunner, s.embedded_tool)
@@ -184,12 +191,8 @@ def tag_git_version(packed):
             packed["http://schema.org/version"] = githash
 
 
-def upload_job_order(arvrunner, name, tool, job_order):
-    """Upload local files referenced in the input object and return updated input
-    object with 'location' updated to the proper keep references.
-    """
-
-    for t in tool.tool["inputs"]:
+def discover_secondary_files(inputs, job_order):
+    for t in inputs:
         def setSecondary(fileobj):
             if isinstance(fileobj, dict) and fileobj.get("class") == "File":
                 if "secondaryFiles" not in fileobj:
@@ -202,6 +205,13 @@ def upload_job_order(arvrunner, name, tool, job_order):
         if shortname(t["id"]) in job_order and t.get("secondaryFiles"):
             setSecondary(job_order[shortname(t["id"])])
 
+def upload_job_order(arvrunner, name, tool, job_order):
+    """Upload local files referenced in the input object and return updated input
+    object with 'location' updated to the proper keep references.
+    """
+
+    discover_secondary_files(tool.tool["inputs"], job_order)
+
     jobmapper = upload_dependencies(arvrunner,
                                     name,
                                     tool.doc_loader,
@@ -286,6 +296,12 @@ class Runner(object):
         self.tool = tool
         self.job_order = job_order
         self.running = False
+        if enable_reuse:
+            # If reuse is permitted by command line arguments but
+            # disabled by the workflow itself, disable it.
+            reuse_req, _ = get_feature(self.tool, "http://arvados.org/cwl#ReuseRequirement")
+            if reuse_req:
+                enable_reuse = reuse_req["enableReuse"]
         self.enable_reuse = enable_reuse
         self.uuid = None
         self.final_output = None